Expert capnogram analysis

A real-time expert system that deciphers CO/sub 2/ waveforms (capnograms) is described. The Capnogram Analyzer Expert System (CAES) was designed using both traditional pattern-recognition methods and an artificial intelligence (AI) approach for signal description and classification. The pattern-recognition technique is used to extract features from the digitized CO/sub 2/ waveforms. The AI approach involves abstracting CO/sub 2/ waveforms from numeric representation to higher-level symbolic representation and a so-called reasoning step to analyze the symbolic data. The CAES consists of three essential components: segmentation, single breath cycle identification and waveform classification. Each component is an expert in itself and is responsible for abstracting the waveform information from a lower level to a higher level using its own domain-specific knowledge base. >